Older adults commonly develop thinking problems and disturbing behavior changes suggesting mild dementia. The challenge confronting clinicians is to determine whether symptoms are due to a progressive dementing illness such as frontotemporal dementia (FTD) or Alzheimer's disease (AD), or a non-dementing psychiatric disorder. Current diagnostic algorithms for such decisions are subjective and based on clinical history and examination, or patients can be observed until the outcome is obvious. Practical, validated diagnostic biomarkers now are urgently needed for more effective and specific treatments. Our group has collected preliminary data suggesting positron emission tomography with SF fluorodeoxyglucose (FDG-PET) is a biomarker that can improve the accurate diagnosis of FTD. We propose a 3-year, prospective, multi-center pilot trial to evaluate in a realistic clinical setting the value of FDG-PET when FTD is suspected. A pilot trial rather than a full-scale study is proposed, because many scientific and logistic advances are necessary first. Nine sites will recruit 6 patients with cognitive impairment and behavioral or language symptoms suggestive of FTD. Two dementia specialists will evaluate each patient using standard clinical methods and make a diagnosis before performing an FDG-PET scan. One then will receive the results of this scan and provide clinical care; the other will remain blinded to these results and perform protocol evaluations every 6 months. After 18 months, a consensus panel with access to all of the patient's data, including autopsy findings in some cases, will determine the final diagnosis. The primary hypothesis of the study is that diagnoses after the FDG-PET scan will be significantly more accurate than those before. In addition, we will determine how FDG-PET results affect physician confidence, patient care and caregiver satisfaction. The consensus panel also will review the outcome of 60 previously studied patients. The results of this study will have broad implications for the development of new biomarkers and the use of FDG-PET in the evaluation of AD and other dementias. In addition to its scientific aims, this pilot trial also will evaluate subject selection and consensus panel procedures and uniform image acquisition and quality control methods needed for the large-scale study.